X64: Fix bugs affecting Win64.

X64: Fix bugs affecting Win64.

Increase stack space on Win64 sample and cctest executables.

[Lasse Reichstein, 2009-10-13 08:40:42]

Please review.

[William Hesse, 2009-10-13 09:15:02]

Some comments.

http://codereview.chromium.org/264047/diff/1/2
File .gitignore (right):

http://codereview.chromium.org/264047/diff/1/2#newcode17
Line 17: d8
Did you really want to make these changes to everyone?  They look good - these
are files we want to ignore.

http://codereview.chromium.org/264047/diff/1/5
File src/execution.h (right):

http://codereview.chromium.org/264047/diff/1/5#newcode222
Line 222: #endif
What is the point of this, if both limits are the same?

http://codereview.chromium.org/264047/diff/1/7
File test/cctest/test-api.cc (right):

http://codereview.chromium.org/264047/diff/1/7#newcode731
Line 731: int32_t value =
Add a comment that the type casts are only to make the compiler happy in the
cases that aren't actually run, due to the guard.  Otherwise, it is confusing -
the type casts actually do the wrong thing, if they do anything.

http://codereview.chromium.org/264047/diff/1/8
File test/cctest/test-assembler-x64.cc (right):

http://codereview.chromium.org/264047/diff/1/8#newcode84
Line 84: #ifdef _WIN64
Why not use a platform dependent macro argument_1_ = rdx or rsi, argument_2_ =
...
Or else static constant register variables arg1 and arg2.

[Lasse Reichstein, 2009-10-13 10:36:21]

http://codereview.chromium.org/264047/diff/1/2
File .gitignore (right):

http://codereview.chromium.org/264047/diff/1/2#newcode17
Line 17: d8
They are either auto-generated files (idp/pdb) or standard temporary files
(i.e., also auto-generated by, e.g., vi and emacs). I see no reason not to have
everyone's git ignore them.

http://codereview.chromium.org/264047/diff/1/5
File src/execution.h (right):

http://codereview.chromium.org/264047/diff/1/5#newcode222
Line 222: #endif
On 2009/10/13 09:15:03, William Hesse wrote:
> What is the point of this, if both limits are the same?

Good point. I'm not sure they should be the same, but for now, I'll remove the
ifdef.

http://codereview.chromium.org/264047/diff/1/7
File test/cctest/test-api.cc (right):

http://codereview.chromium.org/264047/diff/1/7#newcode731
Line 731: int32_t value =
Will do.

http://codereview.chromium.org/264047/diff/1/8
File test/cctest/test-assembler-x64.cc (right):

http://codereview.chromium.org/264047/diff/1/8#newcode84
Line 84: #ifdef _WIN64
A good idea. They should probably be defined somewhere central. I would prefer
the MacroAssembler, but it won't help in the test case, since we test the
Assembler directly.
For now, I'll define it in this file only.

[William Hesse, 2009-10-13 10:42:00]

LGTM.